A startup company says it is bringing a new level to camera-based truck safety systems with the help of artificial intelligence.

NetraDyne was founded last year to apply artificial intelligence to fleet management. It's just announced its first product, driver-i, which uses artificial intelligence to analyze the data coming in through its patent-pending four-camera system and provide near-real-time information on what drivers are doing wrong -- and what they're doing right.

The platform was developed to capture every moment and aspect of the driving experience, rather than a small sample of time.

Driver-i’s artificial intelligence uses deep learning, the approach to AI that most closely mimics the way the human brain processes visual imagery, explained Sandeep Pandya, NetraDyne president.

In addition, the system uses processing technology that was developed for modern smartphones, where the ways they render graphics are similar to deep learning calculations. The system uses the most powerful mobile CPU available, a TerraFlop, which can perform a trillion calculations per second. For comparison, a typical laptop is a GigaFlop, or about a billion per second.

What does all this technology do for the fleet?

“Ultimately this is about replacing the humans having to review videos that come in and try to provide some context later,” explains Adam Kahn, vice president of fleet business. “Our system can do it in real time.”

For instance, say the speed and G-force indicators in the system detect a hard braking event. A traditional camera-based safety system would save a snippet of video before and after the event and send it off to be analyzed by humans, either by a group at the company providing the safety system or by the fleet.

However, Kahn contends, that process A) involves a delay in the information getting to the fleet and B) may not tell the whole story.

He shared an example of a video where a car pulled in front of a truck with a following distance of only 0.4 seconds. The system recognized that it was tailgating, but also realized that it was not the truck driver’s fault.

“It’s a risky event that wasn’t caused by the driver,” Kahn says. “We believe this starts to change the conversation between the fleet manager and the driver.”

“With artificial intelligence we have much greater visibility into the event,” he says, with video of “minutes [not seconds] before and after, with causality determined at the vehicle level so it doesn’t have to go through a center.” There's immediate notification to fleet manager if a severe event happens, he says, and in cases where a driver is being proactive, the fleet manager can provide positive reinforcement and recognition. “I think the more you can have a positive conversation with the driver vs. punitive ones, the more it drives up the ability to retain that driver in your fleet.”

In addition, the artificial intelligence-driven driver-i can capture events that would not be noticed by what he calls “inertial-based events.” For instance, the system detects stop lights and can determine if a driver ran a red light — which would not trigger a traditional safety system, since there’s no hard braking or acceleration or swerving. It also tracks actual truck speed vs. posted limits.

“It’s actually reviewing what is going on every minute of every hour of the driving day,” says Pandya. “It’s always vigilant, it’s always on watch.”

The fleet safety management center is intuitive, full featured and provides near-real time crowd sourcing of driver and vehicle safety event data, says the company, enabling trend analysis and business intelligence with triggers for organization wide work flow.

The driver-i’s proprietary GreenZone Driver Score gives fleets the ability to see how drivers are tracking against team safety programs and goals. This view offers visibility into positive driving activity, events that were caused by third parties, and notification of at-risk conditions, allowing the fleet manager to step in before those events become severe. The GreenZone utility also offers performance-based push notifications so fleet managers can pass along kudos and rewards for excellent driving practices.